Table 6.

Performance Results of 400 Kinship Predictions. The first two rows approximates and summarizes the results obtained in the iDASH 2023 Competition. The third row includes the linear regression solution, which was not submitted to the competition. The fourth row presents an estimate of the naïve solution

Approach	Database encryption	Latency	Throughput	Inferences	auROC	Score
“Unsupervised method” section	8.22 s	14.44 s	27.7 q/s	400	0.96	0.923
“Clustering-based supervised method” section	5.73 s	13.36 s	29.9 q/s	2000	1.0	0.956
“Linear regression method” section	5.73 s	11.63 s	34 q/s	400	1.0	0.962
Naïve	> 5 min	> 10 min	<1 q/m	>800400	1.0	< 0.135

Disclaimer: Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex